跳到正文
DEV Community· Niloy Nath·· 10 小時前AI 評分31

MRRSignal:已驗證 MRR 與自行申報收入,設計讓 AI 解釋而不虛構數字的初創企業資料庫

Verified MRR vs self-reported revenue: designing a startup database where AI explains numbers, never invents them

AI 導讀

MRRSignal 正建立按支付服務供應商驗證的 MRR 排名的公開初創企業收入資料庫,AI 只解釋已驗證數據,不會改動收入數字或排名。平台支援 49 個支付服務供應商,並將收入統一換算為 USD。創辦人可選擇公開列出實際 MRR 或收入區間,是否刊登亦由創辦人決定。

正文

I am building MRRSignal, a public startup revenue database ranked by payment-provider-verified MRR. This is the design problem I keep coming back to: in a product about financial facts, the boundary between a fact and an explanation has to be structural, not a disclaimer in the footer.

A generated paragraph can be useful. It should never quietly become the source of a number.

Start with the source of the revenue figure

Most startup revenue on the internet is self-reported. A founder posts a screenshot or a milestone, and readers decide whether to believe it. MRRSignal moves the source: revenue is read from the founder's connected payment provider through a read-only key, encrypted with AWS KMS. A founder cannot replace the figure with a number typed into a profile.

That makes the origin of the number explicit. It does not make the number self-explanatory - a good page still has to say what metric it shows, what period it covers, and when it was refreshed.

Separate three kinds of output

MRRSignal produces three different things, and the interface treats them differently:

  1. Verified figures - revenue read from the connected payment provider. This is source data.
  2. A statistical revenue projection - a modelled path, clearly an estimate.
  3. A weekly AI outlook - generated text that explains the verified inputs in words.

The rule is simple: nothing an AI writes changes the revenue number or a startup's rank. Projections and outlooks sit downstream of verified data, and the UI labels them so a reader never confuses an estimate with a fact.

Make comparison fair across providers and countries

Startup communities are global, but most revenue tools assume one payment provider and one currency. MRRSignal supports 49 payment providers - global services like Stripe, Braintree, Chargebee and Paddle alongside regional ones such as Razorpay - and normalizes everything to USD. Founders compare on the same scale without being forced onto the same billing stack. Geography should not look like a data-quality problem.

Keep disclosure under founder control

Public transparency is not the same as publishing everything. Listing is opt-in, and founders choose between an exact MRR figure and a revenue band. A band is enough to find peers at your stage without revealing more than you want. Connecting a provider for your own dashboard never silently becomes consent to a public profile.

The question I want feedback on

If you were designing this interface, how would you make source facts, statistical projections and AI-generated explanations visibly different - without burying the user in disclaimers?

I am the founder of MRRSignal. You can see the product at https://mrrsignal.com - it is free to list during beta.

來源:DEV Community · dev.to